RFR (M) CR 8050147: StoreLoad barrier interferes with stack usages

Vladimir Kozlov vladimir.kozlov at oracle.com
Thu Sep 4 08:42:29 UTC 2014


Good.

Thanks,
Vladimir

On 9/4/14 1:39 AM, Aleksey Shipilev wrote:
> Ah, OK.
>
> Let zero to have its own well-deserved "0" then:
>   http://cr.openjdk.java.net/~shade/8050147/webrev.07/
>
> -Aleksey.
>
> On 09/04/2014 12:35 PM, Vladimir Kozlov wrote:
>> Zero is not x86. Otherwise it looks good.
>>
>> Thanks,
>> Vladimir
>>
>> On 9/4/14 1:23 AM, Aleksey Shipilev wrote:
>>> On 09/04/2014 06:28 AM, Vladimir Kozlov wrote:
>>>> Method declaration should be done in src/share/vm/runtime/os.hpp
>>>
>>> Thanks for review, Vladimir!
>>>
>>> I moved the declaration to os.hpp; that allowed me to purge the
>>> duplicate declarations in os_*_*.hpp, moving all implementations to the
>>> appropriate *.cpp. See the updated webrev:
>>>     http://cr.openjdk.java.net/~shade/8050147/webrev.06/
>>>
>>> As usual, passes the JPRT on all platforms.
>>>
>>>> You need make changes in closed sources too.
>>>
>>> Yes. The change is very simple there, I'll post it separately.
>>>
>>> -Aleksey.
>>>
>>>
>
>


More information about the hotspot-compiler-dev mailing list